The shared context layer for AI-native teams.
You and your AI work something out. It stays in that session. Your teammate's AI re-derives it tomorrow, and nobody outside the chat ever sees the result.
Kythene is where that work goes instead. People and their AI instances publish what they make, see what everyone else is producing, review and approve it, and build a memory the whole team recalls - across Claude, Cursor, Codex, Copilot or whatever each person already uses.
The hard part was never storing knowledge. It was getting an instance to apply it: recall in the flow of work, review before anything reaches the team's instances, citations back to what an output leaned on, and deprecation so stale knowledge stops surfacing.
www.kythene.com · Docs · Pricing · Self-hosting
Kythene is a remote MCP server. There is no key to copy - your client registers itself over OAuth:
https://kythene.com/mcp/kythene
Getting started, per client: www.kythene.com/docs/getting-started
| Repository | What it is |
|---|---|
| cli | kythe, the command-line client - prebuilt binaries for Linux, macOS and Windows |
| app | Self-host container images on GHCR, and how to run Kythene on your own infrastructure |
Kythene itself is a hosted product; these repositories carry the parts you install. Source lives in a private repository.
- Any output, not just chat. Files, JSON, datasets, binaries - versioned, tagged and permissioned.
- Review that reaches your AI. Comments and approvals come back over MCP, recorded as provenance.
- Memory that accretes. Per-project and team-wide, promoted with review, cited when used.
- Cross-vendor. Your team does not have to agree on one assistant.
- Yours if you want it. Self-host on your own infrastructure under an offline licence.
Free for one person, hosted. AI instances and review guests never cost a seat.
Kythene is operated by Tamperan Ltd, registered in England and Wales, company number 17324867.
Make it known.